the role of cfd simulation in rocket propulsion support ... · pdf filejuly 26,2011 presented...
Post on 06-Mar-2018
218 Views
Preview:
TRANSCRIPT
July 26,2011
Presented by: Jeff West
Fluid Dynamics Branch (ER42)
Propulsion Systems Department
NASA/Marshall Space Flight Center
MSFC, AL 35812
The Role of CFD Simulation in Rocket Propulsion
Support Activities
https://ntrs.nasa.gov/search.jsp?R=20120016698 2018-05-09T03:26:31+00:00Z
Presentation Overview
CFD at NASA/MSFC
– Flight Projects are the Customer—No Science Experiments
– Customer Support
– Guiding Philosophy and Resource Allocation
– Where is CFD at NASA/MSFC?
Examples of the expanding Role of CFD at NASA/MSFC
– Liquid Rocket Engine Applications
• Evolution from Symmetric and Steady to 3D Unsteady
– Launch Pad Debris Transport-> Launch Pad Induced Environments
• STS and Launch Pad Geometry-steady
• Moving Body Shuttle Launch Simulations
• IOP and Acoustics Simulations
– General Purpose CFD Applications
– Turbomachinery Applications
Parting Thoughts
2
The CFD Customer at NASA/MSFC
NASA/MSFC is a Space Flight Center. This means that Science
Experiments are not appropriate. However, Applied Science to
achieve the Customer’s objectives is desired.
At NASA/MSFC, the Customer is a Program, and most of the time an actual Flight Program.
Examples include:
– Space Shuttle Program
• Space Shuttle Main Engine Element
• Solid Rocket Booster Element
• Ground Operations Element
– Constellation Program
• Ground Operations Element
• Solid Rocket Booster Element
• J-2X Engine Element
• Upper Stage Element
– External Entities
• Air Force
• Commercial Space Flight 3
Definition of Successful Customer Support
In most high visibility situations, CFD must show value to the customer
within three weeks of ATP.
– Once value is demonstrated, the Customer has more patience
The implication of this fact is that the entire CFD simulation process must be optimized for
short turn-around operation
– Geometry and mesh generation
• Mesh generation used to require weeks, now the typical timeframe from CAD
to CFD simulation start is 1 to 2 days.
– Hybrid mix of Commercial and Research software
» ANSA and AFLR (MSU)
– Highly parallel CFD software
• Loci/Chem (MSU) and Loci/STREAM (MSU-SNI)
– High Speed and Low Speed flows, respectively
– Highly parallel post processing software augmented with commercial applications
• Home-grown, batch-mode post processing (AnimatorP)
• Serial commercial software (Fieldview, Ensight, TecPlot)
4
CFD Tool Development Philosophy
Select the most challenging practical and likely future simulation and create a
suite of tools that, when completely developed, can meet that Grand Challenge
– Grand Challenge: Rocket Engine/Motor Combustion Instability
If the Grand Challenge problem is selected properly, then one can achieve
practical progress while also being able to solve the semi-challenging
problems along that path.
– The Grand Challenge tool is then the General Purpose tool.
– Not terribly efficient in terms of lower levels metrics, but efficient at the
most visible level; the time from ATP to first result
The Working Challenge then becomes whether a collection of researchers, CFD
algorithm developers, mesh generation developers and CFD analysts can be
gathered to work in a sustained manner to complete the Grand Challenge.
– This challenge began FY 2001.
5
Resource Allocation Philosophy
Spend controllable monetary resources on growth, not software licenses.
– In FY99, Fluent was $50K/cpu/year…
• For tight budgets, this model fails to allow execution of the CFD
discipline
Growth means software capability and local compute resources
– Support software development with perpetual, unlimited use arrangement
• Required with most government contracts
– Characteristics of efficient local compute resources
• Construct a Linux cluster for one or two applications
• Use open source software
• Choose system administration personnel wisely!
• Dominant cost should be acquisition, not administration
– Exploit remote compute capability as much as possible (NASA/AMES,
DoD, etc)
• User will have virtually no influence over remote compute hardware-
but the taxpayer has already paid for the hardware6
CFD Modeling Branches at MSFC
ER42: Fluid Dynamics Branch, Propulsion Systems Department
– Two Teams performing CFD
• Propellant Delivery Team (14 CFD analysts)
– Propellant Tanks: Slosh and Drain
– Lines Ducts and Valves (MPS and Engine)
– Turbo-machinery (Turbine and Pumps)
– Launch Pad Induced Environments (IOP and Acoustics)
• Combustion CFD Team (8 CFD analysts)
– Lines, Duct, Valves, Manifolds
– Injectors and Thrust Chambers (Performance, Stability and Environments)
– Nozzles (Performance and Environments)
– Solid Rocket Motors (Small and Large Scale)
EV33: Aerosciences Branch, Spacecraft and Vehicle Systems Department
– Two Teams performing CFD
• Aero-Thermodynamics (6 CFD analysts)
– Aerodynamics Heating
– Plume Impingement Heating
– Base Flow Environments
• Aerodynamics Team (4 CFD Analysts)
– Aerodynamic Drag
– Aerodynamics Acoustics 7
Examples of the Expanding Role of CFD at NASA/MSFC:
Liquid Rocket Engine Applications
8
Manifold Flow Uniformity-Steady
2004 to 2007 timeframe
– Steady state simulations
• Primarily Loci/CHEM
– Unstructured mesh simulations
• Gridgen surface geometry, AFLR
Volume mesh
• Integrate to the wall (y+~1)
• 20 to 45 Million cells
– Objective was the assessment of flow
uniformity
9
High Fidelity Manifold Simulations
2007 onwards
• Oxidizer manifold (40 Elements)
– Computational domain include injector inlets
– Loci/Chem and Loci/STREAM
– ANSA Surface Mesh and AFLR Volume
mesh
– Integrate to the wall y+~1
– 30 to 90 Million Cells
• Objectives included Pressure drops in addition to
Flow Uniformity
10
Higher Fidelity Manifold Simulations
11
2007 onwards
• Oxidizer manifold (40 Elements)
– Computational domain include
injector inlets
– Loci/Chem and Loci/Stream
– ANSA Surface Mesh and AFLR
Volume mesh
– Integrate to the wall y+~1
– 30 to 90 Million Cells
• Objectives included Pressure drops in
addition to flow uniformity
Mesh Comparison
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
11.0
12.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0
Chamber Distance (inches)
He
at
Flu
x (
Btu
/in
2-s
)
Penn State CDIT Test Data
Chem-2, SST, Structured, Mesh 1
Chem-3, SST (Pre-Cond), Mesh 2
Chem-3, SST (Pre-Cond), Mesh 3
Chem-3, SST (Pre-Cond), Mesh 4
Chem-3, SST (Pre-Cond), Mesh 5
Loci/Chem and Loci/Stream RANS
―Useful for Design‖
2004-2005
Purdue URANS (CUIP)
―Fully Capable for Design‖
2006-2008
Sandia National Labs LES
―Computational Experiment‖
2006-2008
Rocket Engine Injector Simulations
Penn. State University Validation Experiment Measured Chamber Wall Temperature and Heat Flux.
CFD Simulations Specified Chamber Wall Temperature and Predicted Chamber Wall Heat Flux.
Loci/Chem Axi-symmetric RANS
Purdue Axi-symmetric URANS (CUIP)
Sandia National Labs LES
PSU Shear Coaxial Element
OH-PLIF Image
Rocket Engine Injector Simulations
URANS Axi-symmetric Modeling
Loci/Chem URANS rocket engine injector
simulations have been demonstrated to
reproduce experimental unsteady
characteristics.
14
Loci/Chem
Unsteady 3D Modeling
2009- 2010
• Loci/Stream
• DES
• 40 Mcells
15
Unsteady Duct Modeling
2008-2010
Loci/Chem hybrid RANS-LES simulations
have been applied to improve designs of Liquid
Rocket Engines. Approximately 30 simulations
with 40 to 100 Million Cells.
16
a
a
main inlet
turbine
bypass
duct
main
duct
flow from turbine bypass duct, entrained by main
swirl flow, pours over the inboard side of HEX
flow from turbine
bypass duct, entrained
by main swirl flow,
pours over the inboard
side of HEX
regions of high
total pressure
on HEX engine
inboard side
turbine exhaust
ducting
Examples of the Expanding Role of CFD at NASA/MSFC:
CFD Simulations supporting Shuttle Liftoff Debris Mitigation
17
Space Shuttle Liftoff Debris
As part of the post-Columbia Return to Flight (2003 onwards), the risk of Debris
during the Liftoff Phase was assessed.
– FOD may fall from structure onto SSLV
– FOD/Debris can be propelled upward by plumes
18
Driving Force: Plume-Structure Interactions
To assess plume-driven debris,
accurate predictions of Plume-
Launch Pad Structure
interactions had to be
developed.
19
Steady, Limited Geometry
2004-2005
Significant use and insight
was obtained with steady
state modeling using the
Overflow CFD program.
Plume-structure
interactions remained a
challenge.
20
CFD Model
CAD Model
Steady Trajectory Reconstruction
High Fidelity Geometry- Steady Flow
2005-2007
Conversion to the use of AFLR
unstructured meshes and
Loci/Chem to address the
prediction of upward flows
resulting from Plume-Structure
interactions.
–Aggressive use of
Geometric Symmetry
–Highly detailed
geometry
–25 to 70 Mcells
21
High Fidelity Geometry-Steady Flow
With detailed geometry, the correlation of predicted upward flows to the observed
upward flows increased drastically. Significant insight into the risk posture was
obtained with the analysis and use of the Loci/Chem flowfields. However,
significant uncertainties remained.
– While the CFD accurately predicted the existence of upward flow, the strength and
magnitude of the upward flow did not correlate to an acceptable degree
22
Upward flows contained inside purple isosurfacesPlume Effluent isosurface
Large Domain, High Fidelity Geometry
2007-2009
The high fidelity geometry approach
was extended to the entire Launch Pad.
– 120 to 310 Million Cells
– Three day turn-around
–~50 steady simulations
23
Continued AFLR
development was
instrumental in the
construction of this
model.
Wind-Structure Interactions Resolved
The large-domain,
high-fidelity
geometry
simulations allowed
for the completion
of the risk
assessment due to
gravity-and wind-
driven debris.
24
West Wind
High Fidelity Resolution of All Plumes
Significant
Correlation of
predicted plumes and
plume-structure
interactions was
obtained. However,
there was significant
uncertainty in the risk
due to the steady-
state reconstruction
of the vehicle
trajectory.
25
Simultaneous Plume-Structure Interactions
Remaining
discrepancies
between predicted
and observed
features attributed
to lack of vehicle
motion and the
assumption of
steady state fluid
dynamics.
26
2008 STS-124 Brick Loss Event
CFD Simulations were integral to the assessment of the risk to the Space Shuttle
from Launch Pad Trench Debris if a Trench Failure re-occurred.
27
During the Launch of STS-124, thousands of bricks were
released and ejected from the SRB-side Flame Trench.
Original construction: 1965.
Static Geometry, Transient Simulation
A small computational domain coupled with high-fidelity geometry(30 Mcells)
28
Shuttle Plume Startup Simulation
The Shuttle SRB Startup Simulation resolved fluid dynamic features that are
obscured by the plume during an actual launch
29Shuttle-fountain-effect.mpg
Shuttle Plume Startup Simulation
Observable features were replicated to a good degree.
30Shuttle-mushroom-effect.mpg
Debris Transport Analysis using CFD
The CFD Simulation results allowed the assessment of the transport of Trench debris
as a risk to the Space Shuttle. The program-focused analysis began on 1 June 2008
and was complete by 29 June 2008.
31
Shuttle Moving Body Simulation
2008-2009: Overset Mesh treatment in Loci/Chem coupled with Hybrid RANS-LES allows
accurate resolution of upward flows. 50 to 90 Million Cells.
32
Shuttle-smoke-movingbody.mpg
Shuttle Moving Body Simulation
Shuttle SRB Moving Body with Hybrid RANS/LES treatment also results in more
physically realistic trench flow dynamics.
33
Ares I-X Moving Body Simulation
2009-2010
Overset Mesh treatment in
Loci/Chem coupled with
Hybrid RANS-LES
treatment allows accurate
resolution of plume
Impingement flows and
understanding of Launch
Pad consequences. Two
week turn-around time.
34
Ares-IX-movingbody-hot.mpg
Ares I-X Launch Pad Effects
Very good correlation between CFD –predicted impingement location and
strength with Launch Pad Evidence
35
―What-if?‖ SLS Investigations
Now: The unsteady, moving body capability within Loci/Chem is being used to
evaluate Launch Pad concept consequences for SLS architectures
36SLS-conceptual.mpg
Prediction of Launch Pad Acoustics
Now: Initial Simulations indicate that Loci/CHEM can
predict a useful fraction of Launch Pad Acoustics
37subscale-acoustics-cfd.mpg
Examples of the Expanding Role of CFD at NASA/MSFC:
CFD Simulations supporting Valve Problem Resolution and
Valve Design
38
STS-126 In-flight Valve Failure
2009-2010: The Fuel Flow Control Valves (FCV) supply high-pressure gas from the
SSME engines to pressurize the fuel tank during flight. Three two-state valves are
used to tailor the fuel tank pressure.
39
Unsteady Environments in Static Valves
Loci/Chem was used to model the unsteady loads on the
poppet face using Hybrid RANS-LES capability. The
supersonic annular plume creates unsteady loads on the
poppet face which were never envisoned during the design
and redesign(s) of the FCV. (Root cause was evident to
CFD analysts after first 24 hours on cpu)
40
CFD for SSME Flow Control Valve
• Unsteady simulation of supersonic
impingement on outflow pipe wall
• 15-plus unsteady 3D simulations
with 60 Million cells
• 800 cpus using Pleiades at
NASA/AMES, turn-around in 4 days
Pressure on Poppet Face
Mach Contours in FCVFCV-poppetface.mpg
41
Unsteady CFD SimulationSynthetic Schlierin View
Poppet Face Pressure4 Nodal Diameter Loading Pattern
Supersonic gaseous H2 jet
impinges on Outlet Pipe wall and
pressure waves reverberate
throughout downstream system
Unsteady Environments in Static Valves
42
Unsteady Environments in Static Valves
FCV-pressure-plane.mpg
FCV-mach-plane.mpg
Unsteady Simulation of Valve Transients
Now: Applying the Overset capability of Loci/Chem to simulate valve transient
operation. 120 Mcells, 5 days turn-around, 1600 cpus
43moving-valve-mach.mpg
Examples of the Expanding Role of CFD at NASA/MSFC:
CFD Simulations supporting Turbomachinery
44
Unsteady Simulations of Turbines
2010: Application of the Overset capability of Loci/Chem. One week turn-
around.
45turbine-vortmag.mpg
Multi-Stage Pump Simulations
Now: After porting the Overset capability to Loci/Stream, the unsteady
simulations of pumps is now possible on a large scale. 150-250 Million Cells,
thousands of cpus.
46Pump-multistage.mpg
Parting Thoughts
NASA/MSFC has benefited from the long-term relationship with Mississippi State
University. In addition, MSFC has benefited from the willingness of MSU personnel to
collaborate with entities outside of the University. (Streamline Numerics, Tetra
Research, CFD Research Corporation, ATK, Virginia Tech University, Purdue
University, University of Florida, Sandia National Labs and many others.)
The development and application of new capabilities within both Loci CFD programs
continues to be become more efficient.
– Benefits of the long term collaboration continue to accelerate year after year.
NASA/MSFC’s CFD development path includes Mississippi State University for the
continued advancement of:
– Loci
– Loci/Chem
– Loci/Stream
– AFLR
47
Backup
48
top related